Apparently, it’s blog day!
This post is by Lizzie, and I am requesting analogy help (by the way, thanks for your recent help on how to teach simulation to students).
Yesterday morning I watched a little Metro-Vancouver parks worker trundling along in their tractor, as they gathered up the debris strewn across the beach from our recent storm. The storm had been fantastic fun to cycle home during and I snapped some photos on my ride that do not at all do justice to how riled up the ocean looked (one shown). It also triggered the now-almost-normal stream of requests to link “the severe weather effects we are seeing in [insert place] and how this relates to global warming.”
Which led me to trot out my now very old analogy to explain why we cannot generally attribute any one specific weather event (a specific storm, frost, heat wave etc.) to climate change: consider a MLB player, let’s call her Barry…. For the beginning years of her MLB career she was a pretty good hitter and every so often hit a home run. In the later part of her career she starts taking steroids and hits many more home runs on average. You can’t attribute any particular home run to Barry’s steroid use, but you can associate the changing frequency ….
I didn’t come up with this analogy. I copied it from someone who copied it from someone … and on and on until we find someone who thinks he invented it, but I bet he just forgot where he heard it.
And I like it! People generally get the connection and they are sometimes willing to let go of their urge to pressure me to stay, ‘Whoa! What a storm that was yesterday. That storm was caused by climate change, folks.’ And the steroids fits nicely with our juiced-up climate system so it’s often a good segue into what’s changing in our climate system.
But my analogy feels really out of date! It feels old and I think I lose people who try to remember back when or figure out what I am talking about. I am wondering if anyone has (and is willing to share) a better one they’re using, or wants to propose one I can use.
Research on heat extremes is moving towards terms such as ‘nearly impossible in the absence of warming‘ or ‘virtually impossible without human-caused climate change‘ so maybe I can shelve my example someday? But I am not ready for that. (For anyone waiting on rapid attribution of the PNW storm, I suspect World Weather Attribution is working on it.)
In a small city 10 people a year die of pancreatic cancer. (If you want to add a probabilistic component, make it 10 +- some small random integer. A new factory is built in top of the water supply which leaks chemicals into the water supply. Now 15 (again +- the same small random integer distribution if you like, or increase the variance somewhat) a year die of pancreatic cancer. You can’t attribute any particular death to the chemical leak… indeed, you can be pretty sure that 2/3 of the deaths have nothing to do with the leak at all..
+1 this “attributable risk” example is very helpful
Depends on the size of your hypothetical small city and what fraction of the population 10 or 15 people are.
Still relative increase is 50% (e.g. famous 95% Pfizer efficacy is based on RR, not AR)
It cuts both ways
;-)
I think this example is missing an essential feature. “Was this cancer caused by the chemical leak?” is at least a question that makes sense, since cancers can arise from a single discrete mutation event triggered by a single specific molecule of carcinogen. The problem is just that it is usually difficult for a doctor to determine how the original mutation came about. In principle someone who’d been tracking every cell in the patient’s body could tell you exactly where the cancer came from.
Storms and home runs, on the other hand, are much more complicated in origin. The problem here is not just one of incomplete information. Even if you track every molecule in the atmosphere, it’s still not clear what it would mean for a storm to be “not caused by climate change.” See e.g. https://statmodeling.stat.columbia.edu/2019/09/03/there-is-no-way-to-prove-that-an-extreme-weather-event-either-was-or-was-not-affected-by-global-warming/ A good analogy, like the home-run analogy, should make this feature obvious.
Lizzie:
Well it’s a good thing climate change isn’t working as fast on the earth as steroids did on Barry Bonds.
Here’s if we compare BB’s first five years to his his last five (excluding injury season of 2005):
Seasons, HR Avg, HR Avg Max, Avg # of BB; Avg # of AB
1-5; 0.045; 0.064; 75.4; 520.2
16-21; 0.101, .121; 140.4; 374.6
So in other words, his home run per at bat production doubled in ’16-21 – Even though he was walking almost twice as often! And it’s worse than that, because in the final two seasons of 16-21, HR avg dropped abruptly from about .126 in the first three of those seasons to about .078 in the last two seasons. He retired at *AGE 43*, about half a decade older than the age at which most great hitters lose the quickness to hit (george brett, another great hitter, retired at age 40 after three well below average seasons)
So it’s a good thing the climate in the near term going to go “Peak Bonds”.
Perhaps there should be a metric of at what level of warming a weather event will be ‘Peak Bonds.’ For certain outcomes, I worry we may already have committed to Peak Bonds.
Why not take the analogy out of baseball and into a sport that is more equipment based – a cyclist changes bike, a racing driver changes team, a tennis player changes banana supplier…. Or an athlete changes coach. If they were good before it would be impossible to relate any specific incident to the change, but the overall frequency of performance success would change.
This is only slightly related, but I think it’s worth bringing it around again: https://statmodeling.stat.columbia.edu/2019/09/11/lets-try-this-again-it-is-nonsense-to-say-that-we-dont-know-whether-a-specific-weather-event-was-affected-by-climate-change-its-not-just-wrong-its-nonsensical/
Just clicking back through that post and the post that lead to it and the comments – that was some interesting discussion. Plus it involved a lot of bicycles, which is always good.
My understanding is that in these climate attribution studies they are re-simulating *specific weather events* with and without various drivers as impacted by global warming, and therefore attempting to quantify in what ways and how much anthropogenic climate change contributed. I don’t know the details, but that is my sense of it. So, this is a bit different than Lizzie’s point and your point in that earlier post. They are not just talking about the shifting landscape of weather events, or shifting frequencies, but doing very specific attribution. Somewhat tangentially, I think part of where we run into trouble in most conversations about this topic is the conflict between colloquial use of “affected by” and “caused”.
In that post I wasn’t responding to climate attribution studies, but rather to the the press coverage that one sees all the time, things like “scientists say it’s impossible to attribute any particular weather event to climate change.” If scientists really say that then I think there’s a problem: I think it’s fine to simplify or to use analogies or take other steps to convey science to the general public but I think it’s pretty much always a mistake to say stuff that isn’t true or that doesn’t make sense.
Phili –
> scientists say it’s impossible to attribute any particular weather event to climate change.”
FWIW, I’ve followed this pretty closely.
That’s usually accompanied with a “but scientists do say that climate change likely increases the likelihood of such events occurring, and their severity.”
That messaging occurs as a response to the frequent counter argument of “Severe weather events have occurred prior to AGW, so you can’t say that AGW caused this particular event.”
Looking through your previous post, I notice that you seem to be linking “caused” by global warming and “affected” by global warming. I think that’s a relevant difference. “Skeptics” frequently leverage such a conflation to say that it’s not scientific to argue that a particular storm was “caused” by global warming. Hence, much of the reporting, in response, says “We can’t say this event was ’caused’ by global warming….”
In reply to Phil’s ““scientists say it’s impossible to attribute any particular weather event to climate change.” If scientists really say that then I think there’s a problem…”.
This is exactly what I often say, and it is accompanied by what Joshuar added — “but scientists do say that climate change likely increases the likelihood of such events occurring, and their severity.”
Maybe this is a debate of ’cause’ versus ‘affect’ verbs or what I mean by attribution (which is a pretty particular term in climate change work), but I think it’s an important distinction. There are crazy weather events, now moreso, but also in the past. And we don’t have that much climate data for lots of places where we now see such crazy weather events. And I am asked a lot to say that whatever individual event just happened was caused by climate change, and I think we don’t have great data to link specific events to climate change with that phrasing (and I would not used ‘affected’ either). We understand aspects of the climate system to understand why some events might increase in frequency and/or intensity with climate change, and we can definitely say how much more likely warming has made some events, but I don’t see why, or how, to say specific individual events are climate change affected.
Science is always changing so maybe we’ll get there in a way I don’t see yet (and the late June heat extremes in the PNW may be hitting a new zone in this discussion), but otherwise I don’t think I fully grasp your point (or we’re talking past each other).
Sure, what people care about, and what people probably mean, is what is happening on average: hotter summers on average, more severe storms on average, or whatever.
But, for reasons I outlined in that previous post, I don’t think it makes any sense to say “we can’t say whether this hurricane was caused by climate change” or “we can’t say whether this hurricane was affected by climate change” or anything similar. That hurricane would not have occurred if not for climate change. Maybe some different hurricane would have happened at about the same time at about the same location but if so that is pure coincidence.
In my experience, what statisticians and climatologists have good information on is what happens on average, and what people care about is weather events. I definitely don’t fully get your logic in the previous post; I also think looking a lot a long-term climate data has influenced my view.
Phil, Lizzie:
1. I agree with Lizzie that people (ordinary people and also policymakers) care about extreme events, not about averages. Averages are important because they are more stable to measure than extremes. Estimating averages and then using a model to extrapolate to predict the rate of extreme events can be better, from a statistical perspective, than trying to directly estimate the rate of extreme events.
2. Regarding the causal attribution problem, it would perhaps be simpler to consider an example where the causal mechanism is unquestioned. For example, suppose I shave a pair of dice so that the probability of rolling a 6 on each die increases from 1/6 to 1/4. We then roll the new dice a bunch of times. We can attribute the increased frequency of “boxcars” to the shaving of the dice—but we can’t be sure about any given case. The causal attribution is only probabilistic.
I think Phil is making this point #2, but I feel like the communication may be hindered by being embedded in the controversial example of climate change.
I think that part of this issue relates to socio-pragmatics and communication strategies.
In the past, it was more likely for climate scientists to say that “global warming caused this extreme weather event (e.g., Katrina).”
But the pushback from “skeptics” was on the order of ‘There have been extreme events like this one in the past, so how can you say that global warming ’caused’ this extreme event?”‘
After many years of that pattern repeating, climate scientists began to more regularly add a caveated syntax, that they couldn’t say any single even was “caused” by global warming, but global warming loaded the dice for extreme weather events – making them more frequent and more severe.
In particular when you’re talking about newspaper reporting the comments of climate scientists, this is part of what climate scientists have adapted as a communication strategy. I think it makes sense. I don’t think it’s going to move the needle in terms of convincing “skeptics,” but there’s nonetheless. certain logic to trying to at least consider how to most effectively deliver your message in a way that won’t as easily be dismissed by “skeptics” rhetoric.
Chris –
Not having read you comment, I just wrong something very similar below. It appears to have gotten lost in the ethernet. If it comes back out, apologies for making a similar argument to yours – essentially that the difference between “caused by” and “affected by” is very important in this discussion.
Super old classroom-y metaphor is loaded dice or a card cheat: if they’re good, you don’t notice it happening except by the cheater winning more than their fair share.
You can catch a cold in summer, but it’s more likely in winter. Can you tell which colds specifically are caused by winter? How about in the fall?
Two villages in Bangladesh, one village was encouraged go wear masks, and that village had less Covid infections. Masks protect others, so, if you lived in that village and stayed healthy, was it because of the masks? Many people in the other village stayed healthy as well. (I expect this could prompt some discussion and some thinking.)
An ice cream parlor lowers their prices. On hot days, they now have more customers than before. Can you tell for any individual customer whether the lowered price brought them in? If you ask them, they all say they wanted an ice cream because it’s hot.
Insurance companies sell car insurance based (in part) on the odds of how likely certain car models are to be in an accident. You watch an accident in town and wonder if it’s caused by the car model.
(none of these are tried in conversation, I’m just brainstorming. My feeling is that any situation where you’re increasing the odds of something happening would work, like saving marriages by giving your partner flowers regularly.)
Thanks! I like that many of these are less `negative’ than my MLB example and many others. (The mask one is interesting — communicating one politicized issue with another.)
I’m no help with analogies here, but I like this picture on the blog! What an amazing place to cycle past every day!
It is! If you’re ever in Vancouver, biking along the bay in Kitsilano is pretty amazing.
Great post!
Upon reflecting on the MLB analogy it strikes me that we might be able to say with some confidence which home runs we could attribute to steroid use. In a very simplified model where we assume steroid use makes you hit the ball 10% further you could look deeper into the data and determine which home runs would have been mere long fly balls.
I realize this is stretching the analogy a bit, but it seems it might have an analog in the climate change case! Is there potential to dive deeper into storm data and “control” for climate change?
> changing frequency
Frequency is a property of a collective rather than an individual event.
Back in the 1990’s many statisticians refused to do meta-analysis while routinely analyzing a particular study they were say involved. What surprised me about this, is when I argued in a stats seminar that publication bias does not arise only when you consider the studies together but it is a property of individual studies that only becomes apparent when collected together. Afterwards my mentor told me – “Thanks, now I get why one needs to do meta-analysis. I was always too wary of doing that for the wrong reason.” Again you can’t see bias in an individual study but only in the collection of studies.
In the 1800’s CS Peirce argued that inference was only sensible in collectives and nothing logical could be made of individual events. So it goes back a long time.
I try to make the point various ways
Making sense is formalized in assessments of what would repeatedly happen in some reality or possible world. If it would repeatedly happen (a habit either of an organism, community or physical object/process), it’s real.
The recorded data is of little scientific interest as it just about “dead past” whereas statistic _science_ is directed to the future – what would repeatedly happen in some future.
These concepts are very hard for many to get.